A Residual Dense Network Approach for False Data Injection Attack Localization in Power Grid

False data injection attacks have emerged as a sensitive event in smart power networks with the wide utilization of smart technology-based infrastructure. Hence, these smart power networks are very prone to cyber-physical attacks and sometimes experience system parametric distortions in bus volt-age, phase angle, and power flow measurements. This condition hampers the overall operations considering system security and reliability. Over the years, state estimation has been utilized majorly for detecting false data injection attack events. However, it is observed that attackers have managed to counterfeit the system-measured data and inject improper data into the power network, which results in vulnerabilities in system operations. Earlier research works mainly focus on false data injection event detection. Practically, the exact location should be identified to perform necessary actions to address such unwanted situations. Therefore, this paper presents a deep-learning-based approach, employing a residual dense network (RDN) to address the proper detection of the exact location of the falsely injected data. This work utilizes the IEEE 14 bus data system to perform the test scenarios. The prime objective of employing RDN is to fuse features via identity mapping. The overall findings indicate the elite functioning of the suggested architecture compared to the other available techniques.